Topic: The Alternative Premier League Table 2016-17

[Suareznumber7]

So Arsenal are -10 and we are still -13? United -13 as well I guess now? Spurs need to lose some momentum somehow for us to be able to go top 4. Arsenal always put their shit together once they go out to Bayern in CL.

Right now, It's looking like it'll be one of Liverpool, Utd, Arsenal who will get that 4th spot with the other two fighting it out for 5th/6th.  We need either City or Tottenham to start losing some games. 

[RooiBefok]

APLT Full Table : Round 27

[RooiBefok]

APLT Analysis & Extrapolations : Round 27

[RooiBefok]

LFC Current & Projected Form : Round 27

[JCB]

Relegation APLT: Week 27

This week I've included only the bottom 6 teams in the chart with Watford making an appearance for now. The other teams are included in the table for those who are interested.

[Jim Pooley]

Relegation APLT: Week 27

For comparison, the relegation table under the 'Leicester are shit' model (vs the 'Everton are shit' model above)

Leicester	-6
Swansea	-7
Hull	-9
Palace	-10
Sunderland	-13
Middlesbrough	-14

[JCB]

For comparison, the relegation table under the 'Leicester are shit' model (vs the 'Everton are shit' model above)

Leicester	-6
Swansea	-7
Hull	-9
Palace	-10
Sunderland	-13
Middlesbrough	-14

Interesting... the usual discrepancies as expected for 2 teams swapping pars.
Busy at the moment but wondering it might be worthwhile to check the difference between these models and the actual table (Just to see which is closer - After all the relegation APLT is still in its infancy and it will be a good test for the purposes of assessing and calibrating it).

[Boaty McBoatface]

Chelsea:   +3
Spurs:       -7
Man City:   -9
Arsenal:    -10 (game in hand)
Liverpool:  -13
Man Utd:   -13 (game in hand)

[Prof]

[Suareznumber7]

We need a little help to get top 4 it looks like. 

[sms1986]

Steadying the ship, it seems, which is a good thing for us. I think we'll end up in third or fourth place, either of which will mean great progress!

[SamAteTheRedAcid]

Who are Uniteds games in hand against? Do they have a chance to gain back ALPT points or just par/lose them?

[sms1986]

Who are Uniteds games in hand against? Do they have a chance to gain back ALPT points or just par/lose them?

City and Southampton, I believe. They'll probably lose to City, not sure about Southampton.

[slimbo]

City and Southampton, I believe. They'll probably lose to City, not sure about Southampton.

Both away games

[alvaro]

We need a little help to get top 4 it looks like. 

We dont, the model is wrong. The bookies have us a small favorites to make top 4. 

The big mistake the model does is failing to take into account the difficulty of playing strong teams at home. Arsenal have City and United at home. If they had Burnley and Sunderland instead the model would rate them exactly the same.  Even Liverpool who have been preety excelent at home with 3 wins and 2 draws are -4 in the model.

The other mild mistake the model does is overating how difficult away games against midtable sides are. Liverpool have West Ham away and Stoke City away where the par is 2. None of us would take that point tally if they offered it to us right now even though the model suggests that is title form. In this case this evens out though because Arsenal has two of those games as well.

In Manchester United case they only have one game away against those teams( away at Saints) while they have a hard home game against Chelsea.

So to recap.

Liverpool has two away games against midtable teams giving us a great chance to make ground in the model.
Arsenal has the same two away games which will also give the chance to make ground but they have two really hard par 3 games so it evens out.
Man United have only one away game that will give them a great chance to make ground while they have only one banana skin game.


EDIT: If you want to make the case that Everton is also a banana skin game it doesnt matter because all 3 of us have Everton at home.

EDIT 2: If I was in charge of the model I would make both  home teams 1-6 and away teams 7-12 par 2. In this season the model would be slightly problematic becuase Southampton and Leicester would be overated and Chelsea underated but it would be a failure due to specific circumstances not a structural mistake.

Edit 3: Actually if I was in charge of the model, I would only use the results of the previous season for the first 19 games. After 19 games I would switch to the table at the end of the calendar year and work with that from there on. This small tweak takes care of teams changing their fortunes considerably from one season to another.

[TheShanklyGates]

We dont, the model is wrong. The bookies have us a small favorites to make top 4. 

The big mistake the model does is failing to take into account the difficulty of playing strong teams at home. Arsenal have City and United at home. If they had Burnley and Sunderland instead the model would rate them exactly the same.  Even Liverpool who have been preety excelent at home with 3 wins and 2 draws are -4 in the model.

The other mild mistake the model does is overating how difficult away games against midtable sides are. Liverpool have West Ham away and Stoke City away where the par is 2. None of us would take that point tally if they offered it to us right now even though the model suggests that is title form. In this case this evens out though because Arsenal has two of those games as well.

In Manchester United case they only have one game away against those teams( away at Saints) while they have a hard home game against Chelsea.

So to recap.

Liverpool has two away games against midtable teams giving us a great chance to make ground in the model.
Arsenal has the same two away games which will also give the chance to make ground but they have two really hard par 3 games so it evens out.
Man United have only one away game that will give them a great chance to make ground while they have only one banana skin game.


EDIT: If you want to make the case that Everton is also a banana skin game it doesnt matter because all 3 of us have Everton at home.

EDIT 2: If I was in charge of the model I would make both  home teams 1-6 and away teams 7-12 par 2. In this season the model would be slightly problematic becuase Southampton and Leicester would be overated and Chelsea underated but it would be a failure due to specific circumstances not a structural mistake.

You have to remember that this is a model for winning the league, not for finishing in the top 4. It is based on the presumption that title winning teams tend to win the vast majority of their home games which doesn't always apply to the rest of the top 4.

Par 2s wouldn't make sense since it's impossible to get 2 points from a game at which point the APLT and the actual league table would end up diverging.

[alvaro]

You have to remember that this is a model for winning the league, not for finishing in the top 4. It is based on the presumption that title winning teams tend to win the vast majority of their home games which doesn't always apply to the rest of the top 4.

Par 2s wouldn't make sense since it's impossible to get 2 points from a game at which point the APLT and the actual league table would end up diverging.

The main objective of my post is to make a case of why it's misguided to say Arsenal are clear  favorite over us to finish in the top 4, they are not.

Par 2s are obviously flawed when there is one or two couple games left in the season but I think they are better when evaluating the last 10 games or so of a season. If you look at the West Ham/ Stoke City away games 4 points is more to what a really strong team would be fine with than 2 points. However I understand why people would be wary of this and is highly debatable.
I also think you make a valid point that perhaps the real problem is not the model per se but the fact that we are using a title winning model to evaluate a top 4 race. Anyway whatever the finer points of the discussion I think we can all agree that Arsenal are not 3 point favorite over us at moment.

[me76]

Par 2s??? WTF? Don't mess with the system! It's beautiful!

[RooiBefok]

APLT Full Table : Round 28

[RooiBefok]

APLT Analysis & Extrapolations : Round 28

[RooiBefok]

LFC Current & Projected Form : Round 28

[KurtVerbose]

Hmm, if we win at City, Spurs have Kane injured, Arsenal indifferent form....suddenly it will look very, very close.

If we win at City.

[JCB]

The main objective of my post is to make a case of why it's misguided to say Arsenal are clear  favorite over us to finish in the top 4, they are not.

Par 2s are obviously flawed when there is one or two couple games left in the season but I think they are better when evaluating the last 10 games or so of a season. If you look at the West Ham/ Stoke City away games 4 points is more to what a really strong team would be fine with than 2 points. However I understand why people would be wary of this and is highly debatable.
I also think you make a valid point that perhaps the real problem is not the model per se but the fact that we are using a title winning model to evaluate a top 4 race. Anyway whatever the finer points of the discussion I think we can all agree that Arsenal are not 3 point favorite over us at moment.

I think we all know where you are coming from and can agree to a certain extent with what you say, however trying to fine tune a model can sometimes be counterproductive in the long run.  I did a short course a few years back on machine intelligence and we would feed the system with data and let it try and determine the best model for a data problem (we would use Neural Networks, decision trees, etc - stuff you may have heard in passing when it comes to AI). The algorithms would try hundreds, if not thousands, of different models and assess it's accuracy/prediction capacity relative to one another. The point is there would always end up being an optimal point with every model whereby you needed to find that sweet spot on a curve - Too many rules and it became less accurate, too simple, equally so. In general however it was apparent that simpler models tended to excel more in analysis than one would assume.

Of course each scenario is unique and it could well be that a tweak here or there on the APLT may improve the model. There's no way of knowing without trying...
Harking back to your first post on the subject, if you want to run your ideas and introduce a 'par 2' or run it based on the system at the halfway point my spreadsheet can do most of that for you with minimal work. I've half-arsedly tried both methods in the past with mixed results but that's not to say that you may not be able to do better. As long as your logic is sound and the end results are the same it could well be better.

I'm more than happy to send you a copy if you p.m. me and you can experiment on your ideas at your own leisure. If the results looks convincing (and doesn't complicate things too much) than it can only benefit this thread.

[JCB]

Left it late seeing as the next round of games start tomorrow, Sorry been busy...
Not much of a change but thought I'd include it regardless.

Relegation APLT: Week 28

[KurtVerbose]

Ooh, level with Arsnal.

[ryatnalkar]

Any updates?

[Prof]

[sms1986]

Good, we're stabilising, and I think that shows us the final top four - Chelsea, Spurs, us and Man City.

[Iska]

smh at that Chelsea line.  In my mind they'd reverted to a normal trajectory for their easing home to the title, but they're back on track for the all-time EPL record.

[RooiBefok]

APLT Full Table : Round 29

[RooiBefok]

APLT Analysis & Extrapolations : Round 29

[RooiBefok]

LFC Current & Projected Form : Round 29

[MerseysideBrum]

smh at that Chelsea line.  In my mind they'd reverted to a normal trajectory for their easing home to the title, but they're back on track for the all-time EPL record.

Kinda glad really, if would of come down to near everyone else it would of just made me more angry at our mishaps. At least this way it was practically impossible for us ever to of beaten them to the title anyway so it's easier to swallow

[Kochevnik]

So today's games leave us with the following if my maths are correct:

Chelsea: P29 +2
Spurs: P29 -7
Man City: P28 -11
Liverpool: P30  -13
Arsenal: P27  -13
United: P28  -15

Tomorrow's Arsenal/City game is a big one for the APLT.  It's a Par 3 game for Arsenal, Par 1 for City.

[sms1986]

So today's games leave us with the following if my maths are correct:

Chelsea: P29 +2
Spurs: P29 -7
Man City: P28 -11
Liverpool: P30  -13
Arsenal: P27  -13
United: P28  -15

Tomorrow's Arsenal/City game is a big one for the APLT.  It's a Par 3 game for Arsenal, Par 1 for City.

Looks very similar to how I think the top six will end up, although I think we can still get third if not second (and maybe even first as it is mathematically possible, but yeah I know it's almost impossible in reality).

[drmick]

Relegation APLT: Week 27

This week I've included only the bottom 6 teams in the chart with Watford making an appearance for now. The other teams are included in the table for those who are interested.

After Palace's win yesterday, there is now a clear bottom two. That 3rd spot will be very interesting. It may be worthing putting some of the teams back in because of that tightness. The table only gives the status at the moment, but the graph shows a trajectory of form (like the improvement in Leicester and Palace).

[supaerheraw]

Looks very similar to how I think the top six will end up, although I think we can still get third if not second (and maybe even first as it is mathematically possible, but yeah I know it's almost impossible in reality).

Yeah, you never know in football. They've got some dodgy games up ahead. Thinking logically Chelsea will at maximum drop 5 points in the APLT, putting them at -3 which won't be enough for us even if we win all of the remaining games. Anything more than that and we're looking at a Chelsea in free fall much like us a couple of months back — which... isn't too likely.

Thinking about the remaining games I'd put my money on us finishing third. But then again, that's if our team doesn't decide to take another dive. 

I believe this is the year Arsenal falls out of top four. They've got a nasty run in.

[Pie Eyed]

Well, it's looking like the perfect result so far.

The draw is best all round, in terms of simple numbers.

Real table:
If City Win, Arsenal drop 3 points against us
If Arsenal win, City drop 3 points against us.
If it's a draw, 4 points are lost against us in total (2 each).

Same with the APLT:
City Win, they gain 2 but Arsenal lose 3, total = -1
Arsenal win, they get 0, City lose 1, total = -1
Draw, City get 0, Arsenal lose 2, total = -2

Sorry for being a bit convoluted, but I couldn't decide who I wanted to lose more, so it all came down to the numbers.   

[GIPPO77]

Draw... probably best for us.
 

[Prof]